Actions, Reasons, Neurons and Causes

1
Actions, Reasons, Neurons and Causes Jeffrey D.
Schall Narcisse Bichot, Leanne Boucher, Josh
Brown, Corrie Camalier, Jeremiah Cohen, Erik
Emeric, Doug Hanes, Richard Heitz, Shigehiko Ito,
Chi-Hung Juan, Min-Suk Kang, Aditya Murthy,
Matthew Nelson, Pierre Pouget, Chenchal Rao,
Supriya Ray, Takashi Sato, Stephanie
Shorter-Jacobi, Veit Stuphorn, Tracy Taylor,
Kirk Thompson Geoff Woodman with Jean
Bullier, Jon Kaas, Audie Leventhal, Gordon
Logan, Anne Morel, Tom Palmeri, Andrew Rossi
Workshop 7 Systems Biology of Decision
Making Mathematical Biosciences Institute June
2008
2
We are spending time on details, but I want to
adopt broader perspective. Consider (1) The
bee decided to fly North. (2) Bush decided to
invade Iraq. Does (2) include something more
than (1)? If not, what will become of law
human relations? If so, what? The answer, it
seems to me, depends on understanding what we
mean by decide.
3
Definitions

• Choice action in the context of alternatives
  to satisfy a goal, desire or preference
• coffee or tea?
• choices take time

• Action anything we do
• Actions have reasons - I did
• Events just have causes - It happened
• Reasons for actions are explanations in terms of
  purposes, i.e., intentions
• A particular movement may be intentional under
  one description but not under another
• e.g., a wink or a blink
• Decision deliberation when alternatives vague,
  payoffs unclear or habits reversed
• New Guinea Peaberry or Bella Vista F.W. Tres
  Rios Costa Rica?

4
Characteristics of decision
Unlike choices, decisions cannot be predicted.
The source of decisions is inaccessible to
introspection.
"I feel that way right now. Ask me in two or
three months and I may change. I don't think I
will. I'm pretty sure that's my decision."
Michael Jordan on his retirement from
professional basketball. Associated Press, 17
July 1998
I look forward to playing and hopefully I can
get to that point where I can make that
decision. Michael Jordan on his anticipated
return to professional basketball. Associated
Press, 19 July 2001
5
Refining definition of decision

• Distinguish two meanings
• (1) As quantitative rules describing behavior
  (Game theory, Economics)
• But average measures of outcome do not specify
  mechanism
• (2) As process producing behavior
• Mechanism with a particular architecture
• Plausible mechanisms can be described
  mathematically, e.g., signal detection theory,
  drift diffusion, EBRW, ITAM, TVA
• Decision as process has two distinct meanings
• (1) Decide to
• Alternative actions
• Can be identified with choosing
• Good/bad but not true/false
• (2) Decide that
• Alternative categories
• True/false

6
An empirical basis for distinguishing between
choosing and deciding
It is deciding when medial frontal cortex is
engaged.
Rees et al. Nature Neuroscience 3, 716 - 723
(2000)
7
Necessity of formal linking propositions

• The properties of neurons do not reveal function
  
• Formal (computational) theories of performance
  explain function
• But distinct models cannot be distinguished from
  behavior testing, e.g., diffusion or race
• Properties of neurons might provide constraints
  to distinguish between models
• if and only if the neural activity measured is
  the instantiation of the cognitive process in
  question, which constitutes a linking proposition

Teller DY. 1984. Vision Research
241233-1246 Schall JD. 2004. Ann Rev Psychol
5523-50
8
Linking propositions for decision making
Hanes Schall (1996) described neural activity
that looked like an accumulator.
9
Linking propositions for decision making
RT Decision time Residual time Residual time
Encoding time Preparation time
10
An experimental system
How does the brain choose where to look?
How does the brain control when to move?
How does the brain correct errors?
11
Saccade target selection and preparation are
accomplished by a distributed network in the brain
Munoz DP, Schall JD (2003) Concurrent distributed
control of saccade initiation in the frontal eye
field and superior colliculus. In The Oculomotor
System New Approaches for Studying Sensorimotor
Integration. Edited by WC Hall, AK Moschovakis.
CRC Press, Boca Raton, FL. Pages 55-82.
12
How the brain chooses where to look
13
How the brain chooses where to look
Correct
Error
14
Choosing target and choosing saccade
Murthy A, Thompson KG, Schall JD. (2001) Dynamic
dissociation of visual selection from saccade
program-ming in frontal eye field. J
Neurophysiol. 862634
15
Control of responses investigated with stop
signal task
16
Countermanding performance
Probability of not canceling increases with stop
signal delay
Probability (Not cancel)
i
l
i
b
Pr
17
Countermanding paradigm Race model
GO
GO
STOP
Logan, G.D. Cowan, W.B. (1984) On the ability
to inhibit thought and action A theory of an act
of control. Psychological Review 91295-327.
Hanes DP and Schall JD (1995) Countermanding
saccades in macaque.Visual Neuroscience 12929-937
18
Saccades are produced by a distributed network
Munoz DP, Schall JD (2003) Concurrent distributed
control of saccade initiation in the frontal eye
field and superior colliculus. In The Oculomotor
System New Approaches for Studying Sensorimotor
Integration. Edited by WC Hall, AK Moschovakis.
CRC Press, Boca Raton, FL. Pages 55-82.
19
Origin of response time variability
Neural Activity
0.0
0.1
0.2
Time from stimulus (sec)
Hanes, D.P. and J.D. Schall (1996) Neural control
of voluntary movement initiation. Science
274427-430.
20
Countermanding physiology
Non-canceled saccades occur when movement-related
activity reaches the threshold before SSRT
STOP
SSRT
Hanes, D.P., W.F. Patterson, J.D. Schall (1998)
The role of frontal eye field in countermanding
saccades Visual, movement and fixation activity.
Journal of Neurophysiology 79817-834. Pare M,
Hanes DP (2003) Controlled movement processing
superior colliculus activity associated with
countermanded saccades. Journal of Neuroscience
236480-6489.
21
Fixation cells in FEF SC contribute to stopping
saccades
SSRT
Stop Signal
SSRT
Stop Signal
Activation (Spikes/sec)
100
400
200
0
400
200
0
Time from target (ms)
Time from target (ms)
Tempting to believe that movement cells are
inhibited by fixation cells.
Hanes, D.P., W.F. Patterson, J.D. Schall (1998)
The role of frontal eye field in countermanding
saccades Visual, movement and fixation activity.
Journal of Neurophysiology 79817-834.
Paré M, Hanes DP (2003) Controlled movement
processing superior colliculus activity
associated with countermanded saccades. Journal
of Neuroscience 236480-6489.
22
Mapping the race model onto neural processes
two facts expose a paradox
1 - The race model of countermanding performance
assumes that the GO and the STOP processes have
independent finish times.
2 Saccades are produced by a network of
interacting neurons.
How can a network of interacting neurons produce
behavior that looks like the outcome of a race
between independent processes?
23
The STOP unit must inhibit the GO unit. But, if
the inhibition is uniform and instantaneous, then
the non-canceled movement will have longer than
observed initiation times. Therefore, the
inhibition of the STOP unit on the GO unit must
be late and potent.
24
Delayed potent STOP
STOP
SSRT
Boucher L, Logan GD, Palmeri TJ, Schall JD (2007)
Inhibitory control in mind and brain An
interactive race model of countermanding
saccades. Psychological Review 114376-397
25
Delayed potent STOP
Reproduces countermanding behavior
Boucher L, Logan GD, Palmeri TJ, Schall JD (2007)
Inhibitory control in mind and brain An
interactive race model of countermanding
saccades. Psychological Review 114376-397
26
Delayed potent STOP
and reproduces neural activation
The GO unit is not modulated in non-canceled
trials
Boucher L, Logan GD, Palmeri TJ, Schall JD (2007)
Inhibitory control in mind and brain An
interactive race model of countermanding
saccades. Psychological Review 114376-397
27
But what about errors?
The medial frontal lobe monitors consequences and
conflict.
28
Error signals from single neurons correspond to
scalp potentials
and are observed in ACC of monkeys.
Non-canceled error trials
Correct no stop signal trials
Emeric EE, Brown JW, Leslie M, Pouget P, Stuphorn
V, Schall JD (2008) Error-related local field
potentials in the medial frontal cortex of
primates. Journal of Neurophysiology 99759-772.
29
A framework for understanding guidance control
of action.
30
What is the link between neural causes and
intentional reasons? For lower animals and
reflexes the mapping of brain state to behavior
is one-to-one
(What about bees and ants and fish individually
and as hives, colonies and schools?)
31
Many-to-one mapping

• The same eye movement can originate from
  different brain states
• An eye movement of a given direction can be
  evoked by activation of a particular site in the
  superior colliculus or frontal eye field

32
One-to-many mapping

• Different movements can occur based on a single
  representation of the world

• If the brain knew where the target was, why
  did it make an error?
• Why do you say things you dont mean? The mouth
  moves faster than the mind

33
Reconciling intentional reasons with neural causes

• If a given body movement can arise from
  different brain states, then the dependence of
  behavior on intention can be explained in terms
  of the representational content of the intention
  (reasons) and not its neural realization as such
  (causes)
• A movement can be called an intentional action
  if and only if it originates from a cognitive
  state with meaningful content which is the reason
  for the action
• The representation of a single focus of
  activation in the brain leading to an eye
  movement of a particular direction can be
  distinguished from the representation of two foci
  of activation leading to the same saccade through
  averaging.
• But, the two mappings of neural representations
  onto saccades do not have equal status.
• Averaging eye movements are maladaptive
  because they direct gaze to neither stimulus
  they are unintentional errors that must be
  corrected to achieve the goal of vision.
• In contrast, an accurate saccade to one of the
  two stimuli would achieve the goal of vision and
  more likely would be owned as intentional.
• Self-monitoring distinguishes I did from it
  happened

34
(No Transcript)